Research Topic

Our research topic is Trend of Major Types of Crimes commited by White Males in the DC Area in 2016-2021. We chose this topic because we are interested in the impact of COVID-19 on crimes. We will use the data provided by the Metropolitan Police of DC regarding adult arrests over a time period stretching between 2016-2021.

Data Shaping

Read Data

We read the data .CSV files of adults arrest in DC area from 2016-2021

The column names of the data in 2016 and 2017 were not the same with others. The below table shows the column names of the data in 2016 and the data in 2016.

col # 2016 2018
1 Arrestee.Type Arrestee.Type
2 Arrest.Year Arrest.Year
3 Arrest.Date Arrest.Date
4 Arrest.Hour Arrest.Hour
5 CCN CCN
6 Arrest.Number. Arrest.Number.
7 Age Age
8 Defendant.PSA Defendant.PSA
9 Defendant.District Defendant.District
10 Defendant.Race Defendant.Race
11 Defendant.Ethnicity Defendant.Ethnicity
12 Defendant.Sex Defendant.Sex
13 Arrest.Category Arrest.Category
14 Charge.Description Charge.Description
15 Arrest.Location.PSA Arrest.Location.PSA
16 Arrest.Location.District Arrest.Location.District
17 Arrest.Location.Block.GeoX Arrest.Block.GEOX
18 Arrest.Location.Block.GeoY Arrest.Block.GEOY
19 Offense.GEOY Arrest.Latitude
20 Offense.GEOX Arrest.Longitude
21 Offense.PSA Offense.Location.PSA
22 Offense.District Offense.Location.District
23 Arrest.Latitude Offense.Block.GEOX
24 Arrest.Longitude Offense.Block.GEOY
25 Offense.Latitude Offense.Latitude
26 Offense.Longitude Offense.Longitude

Merge Multiple CSV Files and Drop Some Columns

The column names were same from the first column to the 14th column in both data. On the other hand, the name and order of 15th and latter columns were a bit different in those data. The latter columns were about locations, and we were not very interested in the detail location. Therefore, we deleted the latter columns except for the 16th and 22nd columns. In addition, we dropped CNN (col #5) and Arrest.Number. (col #6) because they were IDs and useless for our analysis.

The format of date was different from years; the data in 2016 and 2017 has the format like , the data in 2018 to 2020 has the format like , and the data in 2021 has the format like . We coverted Since different date formats for different years are difficult to analyze, we will unify the date format to “yyyy-mm-dd”.

After deleting some columns and changing the date format, we binded data frames by rows.

Correct Anomalies

Remove abnormal values

To see whether there were abnormal values, we created the table showing some statistics for numerical variables.

Table: Statistics summary.
Arrest.Year Arrest.Hour Age
Min Min. :2016 Min. : 0.00 Min. : 18.00
Q1 1st Qu.:2017 1st Qu.: 6.00 1st Qu.: 25.00
Median Median :2018 Median :12.00 Median : 32.00
Mean Mean :2018 Mean :11.81 Mean : 35.19
Q3 3rd Qu.:2019 3rd Qu.:18.00 3rd Qu.: 43.00
Max Max. :2021 Max. :23.00 Max. :121.00

The maximum age was too old. 55 rows were assigned an age of over 100 years (117-121 ) in these data, and it seemed to be wrong. Therefore, we dropped these rows.

Some data cleaning.. Dropping, Binding and Renaming of columns as needed.

##  [1] "Arrestee_Type"             "Arrest_Year"              
##  [3] "Arrest_Date"               "Arrest_Hour"              
##  [5] "Age"                       "Defendant_PSA"            
##  [7] "Defendant_District"        "Defendant_Race"           
##  [9] "Defendant_Ethnicity"       "Defendant_Sex"            
## [11] "Arrest_Category"           "Charge_Description"       
## [13] "Arrest_Location_District"  "Offense_Location_District"

Missing Values

##             Arrestee_Type               Arrest_Year               Arrest_Date 
##                         0                         0                         0 
##               Arrest_Hour                       Age             Defendant_PSA 
##                         0                         0                     29093 
##        Defendant_District            Defendant_Race       Defendant_Ethnicity 
##                      9337                         0                         0 
##             Defendant_Sex           Arrest_Category        Charge_Description 
##                         0                        12                        15 
##  Arrest_Location_District Offense_Location_District 
##                       184                        11
Our variables of concern in thsi dataset contain rich data.. over 95% of data available in each useful column.. will not delete any rows
##  [1] "Arrestee_Type"             "Arrest_Year"              
##  [3] "Arrest_Date"               "Month"                    
##  [5] "Day"                       "Arrest_Hour"              
##  [7] "Age"                       "Defendant_PSA"            
##  [9] "Defendant_District"        "Defendant_Race"           
## [11] "Defendant_Ethnicity"       "Defendant_Sex"            
## [13] "Arrest_Category"           "Charge_Description"       
## [15] "Arrest_Location_District"  "Offense_Location_District"
## Warning: 程辑包'lubridate'是用R版本4.2.2 来建造的
## 
## 载入程辑包:'lubridate'
## The following objects are masked from 'package:base':
## 
##     date, intersect, setdiff, union

Correct inconsistent values

Arrest_Category had some different values for 2021 and other years:

  • Data in 2021 had “Release Violations/Fugitive (Fug)” and “Release Violations/Fugitive (Warr)” although data in other years have “Release Violations/Fugitive” instead of them.
  • Data in 2021 had “Fraud and Financial Crimes (Frau)” although data in other years have “Fraud and Financial Crimes”.

Therefore, we coverted these values in 2021 into the correspond values in other years.

Remove Unnecessary Rows

Since we were interested in crimes committed by while males, we dropped rows where the value of Defendant_Race was not “White”. The structure of the final data is shown in the below table.

Data frame structure
column_name class first_values
Arrestee_Type character Adult Arrest, Adult Arrest, Adult Arrest, Adult Arrest, Adult Arrest, Adult Arrest
Arrest_Year integer 2016, 2016, 2016, 2016, 2016, 2016
Arrest_Date double 2016-01-01, 2016-01-01, 2016-01-01, 2016-01-01, 2016-01-01, 2016-01-01
Month integer 01, 01, 01, 01, 01, 01
Day integer 1, 1, 1, 1, 1, 1
Arrest_Hour integer 0, 0, 1, 1, 13, 2
Age integer 39, 27, 27, 26, 48, 25
Defendant_PSA character Out of State, Out of State, Out of State, Out of State, 404, Out of State
Defendant_District character Out of State, Out of State, Out of State, Out of State, 4D, Out of State
Defendant_Race integer WHITE, WHITE, WHITE, WHITE, WHITE, WHITE
Defendant_Ethnicity character UNKNOWN, NOT HISPANIC, HISPANIC, NOT HISPANIC, NOT HISPANIC, HISPANIC
Defendant_Sex integer MALE, MALE, MALE, MALE, MALE, MALE
Arrest_Category character Simple Assault, Simple Assault, Driving/Boating While Intoxicated, Simple Assault, Simple Assault, Simple Assault
Charge_Description character Threats To Do Bodily Harm -misd, Simple Assault, Driving While Intoxicated -2nd Off, Simple Assault, Simple Assault, Simple Assault
Arrest_Location_District integer 2D, 3D, 4D, 5D, 1D, 3D
Offense_Location_District integer 2D, 3D, 4D, 5D, 1D, 3D
Weekday integer 星期五, 星期五, 星期五, 星期五, 星期五, 星期五

EDA

Crimes Committed by All

Hour of the day.. at what time do these crimes occur the most?

## # A tibble: 24 × 2
##    Arrest_Hour Total
##          <int> <int>
##  1           0  5681
##  2           1  7425
##  3           2  6769
##  4           3  6286
##  5           4  5425
##  6           5  4412
##  7           6  4465
##  8           7  6101
##  9           8  6729
## 10           9  6594
## # … with 14 more rows

Day .. What day of the month has the highest crime incidents?

## # A tibble: 31 × 2
##      Day Total
##    <int> <int>
##  1     1  5516
##  2     2  5193
##  3     3  5240
##  4     4  5138
##  5     5  5189
##  6     6  4998
##  7     7  4878
##  8     8  5022
##  9     9  4947
## 10    10  5090
## # … with 21 more rows

Day of the week..

## # A tibble: 7 × 3
##   Weekday Total Percent
##   <fct>   <int>   <dbl>
## 1 星期一  19553    12.8
## 2 星期二  21418    14.1
## 3 星期三  23520    15.4
## 4 星期四  23241    15.3
## 5 星期五  23189    15.2
## 6 星期六  22142    14.5
## 7 星期日  19268    12.6

Monthly crime incidence

## # A tibble: 12 × 3
##    Month Total Percent
##    <fct> <int>   <dbl>
##  1 01    12751    8.37
##  2 02    12158    7.98
##  3 03    13625    8.94
##  4 04    12344    8.10
##  5 05    13427    8.81
##  6 06    12729    8.36
##  7 07    13008    8.54
##  8 08    12991    8.53
##  9 09    12578    8.26
## 10 10    13029    8.55
## 11 11    11869    7.79
## 12 12    11822    7.76

Crime incidence grouped into yearly plots..

## # A tibble: 6 × 3
##   Arrest_Year Total Percent
##   <fct>       <int>   <dbl>
## 1 2016        29980    19.7
## 2 2017        31209    20.5
## 3 2018        29100    19.1
## 4 2019        27915    18.3
## 5 2020        18479    12.1
## 6 2021        15648    10.3

Some Boxplots

This plot is to compare the age of people who get arrested with each different year. From the plot, we can see that there are lots of outliers. We need to get rid of the outliers first.

## [1] 151552     17
## [1] 152331     17

Removed 834 outliers.

After remove the ourliers, we could clearly see that as year goes up, the minimum age goes up a little bit. The maximum age from this sample goes down. The median is pretty much same compare to different years.

There are less younger criminals as year passing from 2016 to 2020 based on this sample. Criminal with age from 30 to 35 arrested more than other ages, which also probably means that there are more offenders with age from 30 to 35.

Why boxplot? The advantage of consider median over sample mean is that it is less affected by extreme observations.

## [1] WHITE    BLACK    UNKNOWN  ASIAN    MULTIPLE OTHER   
## Levels: ASIAN BLACK MULTIPLE OTHER UNKNOWN WHITE
## [1] 15739
## [1] 131385
## [1] 897
##  [1] "Simple Assault"                     "Assault on a Police Officer"       
##  [3] "Traffic Violations"                 "Weapon Violations"                 
##  [5] "Driving/Boating While Intoxicated"  "Narcotics"                         
##  [7] "Disorderly Conduct"                 "Theft"                             
##  [9] "Liquor Law Violations"              "Other Crimes"                      
## [11] "Theft from Auto"                    "Offenses Against Family & Children"
## [13] "Assault with a Dangerous Weapon"    "Release Violations/Fugitive"       
## [15] "Motor Vehicle Theft"                "Damage to Property"                
## [17] "Sex Abuse"                          "Property Crimes"                   
## [19] "Vending Violations"                 "Robbery"                           
## [21] "Aggravated Assault"                 "Burglary"                          
## [23] "Sex Offenses"                       "Fraud and Financial Crimes"        
## [25] "Prostitution"                       "Homicide"                          
## [27] "Kidnapping"                         "Gambling"                          
## [29] "Arson"                              NA

From the box plot, we can see that most sex abuse happens around 10am to 13pm. The sex abuse happens all the time and it changes with different years.

From the box plot, we can see that most theft happens around 15pm and they all super same with each year except year 2019. The theft always happening from 11am to 19 pm. That’s a funny fact.

Crimes Committed by White Male

Sex distribution Race distribution From the barplot of Sex, we can see there is not enough female samples. From the barplot of Race, there are too many black people and to small other races which is not appropriate for us to do analysis. Therefore, we choose to investigate in white male crimes.

same pattern here as above, will dig into a few other stuff too..

White Men - By the Hour

White Men - What day of the month has the highest crime incidents?

## # A tibble: 31 × 2
##      Day Total
##    <int> <int>
##  1     1   524
##  2     2   404
##  3     3   403
##  4     4   395
##  5     5   405
##  6     6   455
##  7     7   405
##  8     8   376
##  9     9   399
## 10    10   430
## # … with 21 more rows

White Men - What Day of the Week has the highest crime incidents?

## # A tibble: 7 × 3
##   Weekday Total Percent
##   <fct>   <int>   <dbl>
## 1 星期一   1566    12.8
## 2 星期二   1463    11.9
## 3 星期三   1639    13.3
## 4 星期四   1736    14.1
## 5 星期五   1876    15.3
## 6 星期六   2046    16.7
## 7 星期日   1952    15.9

White Males - By Month

## # A tibble: 12 × 3
##    Month Total Percent
##    <fct> <int>   <dbl>
##  1 01     1189    9.68
##  2 02      998    8.13
##  3 03     1125    9.16
##  4 04      921    7.50
##  5 05     1058    8.62
##  6 06     1010    8.23
##  7 07      966    7.87
##  8 08      961    7.83
##  9 09     1023    8.33
## 10 10     1108    9.02
## 11 11      991    8.07
## 12 12      928    7.56

White Men - Yearly Crime Incidents

## # A tibble: 6 × 3
##   Arrest_Year Total Percent
##   <fct>       <int>   <dbl>
## 1 2016         2620   21.3 
## 2 2017         2636   21.5 
## 3 2018         2297   18.7 
## 4 2019         2191   17.8 
## 5 2020         1425   11.6 
## 6 2021         1109    9.03

Let us get into crime types…

## # A tibble: 10 × 2
##    Arrest_Category                   Total
##    <chr>                             <int>
##  1 Simple Assault                     2661
##  2 Traffic Violations                 1548
##  3 Release Violations/Fugitive        1133
##  4 Driving/Boating While Intoxicated  1045
##  5 Other Crimes                        821
##  6 Theft                               674
##  7 Narcotics                           654
##  8 Liquor Law Violations               562
##  9 Disorderly Conduct                  433
## 10 Damage to Property                  414

## `summarise()` has grouped output by 'Arrest_Year'. You can override using the
## `.groups` argument.
## # A tibble: 10 × 3
## # Groups:   Arrest_Year [1]
##    Arrest_Year Arrest_Category                   Total
##    <fct>       <chr>                             <int>
##  1 2016        Aggravated Assault                   23
##  2 2016        Assault on a Police Officer          42
##  3 2016        Assault with a Dangerous Weapon      73
##  4 2016        Burglary                             25
##  5 2016        Damage to Property                   98
##  6 2016        Disorderly Conduct                   83
##  7 2016        Driving/Boating While Intoxicated   206
##  8 2016        Fraud and Financial Crimes           11
##  9 2016        Homicide                              2
## 10 2016        Kidnapping                            4

Some Top Crimes

We see there are some wired high peaks in the count plots above. This is due to some protests in DC area which caused the surge of crimes. If we remove these peaks, they will be simliar to other plots, but we cannot, these are related to the events happened in the real world.

Location Stuff

##  [1] 2D      3D      4D      5D      1D      7D      6D      UNKNOWN <NA>   
## [10]        
## Levels:  1D 2D 3D 4D 5D 6D 7D UNKNOWN
## 
##              1D      2D      3D      4D      5D      6D      7D UNKNOWN 
##      22    1858    3053    2555    2816    1231     437     260      29
##  [1] 2D      3D      4D      5D      1D      7D      6D      #N/A    UNKNOWN
## [10] Unk    
## Levels: #N/A 1D 2D 3D 4D 5D 6D 7D Unk UNKNOWN
## 
##    #N/A      1D      2D      3D      4D      5D      6D      7D     Unk UNKNOWN 
##      12    2044    3100    2531    2719    1195     413     238      16      10

## `summarise()` has grouped output by 'Arrest_Location_District'. You can
## override using the `.groups` argument.
## Selecting by Total

## `summarise()` has grouped output by 'Offense_Location_District'. You can
## override using the `.groups` argument.
## Selecting by Total

Number of Each Crime

We created some bar plots to see the number of occurrences per type of crime.
The Bar plots of crimes in each year are as follows:

“Offenses Against Family & Children” have been increasing after COVD-19.

The top 6 crimes (or 7 crimes when ‘Other Crimes’ are included) in each year are as follows.

Rank 2016 2017 2018 2019 2020 2021
1 Simple Assault Simple Assault Simple Assault Simple Assault Simple Assault Simple Assault
2 Traffic Violations Traffic Violations Traffic Violations Traffic Violations Driving/Boating While Intoxicated Traffic Violations
3 Release Violations/Fugitive Release Violations/Fugitive Release Violations/Fugitive Prostitution Release Violations/Fugitive Driving/Boating While Intoxicated
4 Driving/Boating While Intoxicated Driving/Boating While Intoxicated Driving/Boating While Intoxicated Driving/Boating While Intoxicated Traffic Violations Release Violations/Fugitive
5 Liquor Law Violations Other Crimes Narcotics Release Violations/Fugitive Offenses Against Family & Children Other Crimes
6 Narcotics Disorderly Conduct Theft Other Crimes Other Crimes Offenses Against Family & Children
7 NA Liquor Law Violations NA Theft Narcotics Damage to Property

Time Series Change in the Number of Major Crimes

To see the trend of the above major crimes, we created a line plot as follows.

“Simple Assault”, “Traffic Violations”, and “Theft” have clearly declined since 2020. On the other, “Offenses Against Family & Children” has increased in 2020 and 2021 compared to previous years. COVID-19 seems to be related to these trend change. We posed the following SMART QUESTION, and we will analyze these four crimes in detail in the following.

Is there a significant difference in “Simple Assault”, “Traffic Violations”, “Theft”, and “Offenses Against Family & Children” trends among adult white males within the DC area between 2016 and 2021, and could COVID protocols play a role in these trend shifts?

Common Crimes trend among people with different generations

group white males by age:

Young: 18-29; Middle age: 30-50; Old: >50. Crimes:“Simple Assault”,“Release Violations/Fugitive”,“Traffic Violations”, “Narcotics”, “Theft”, “Other Crimes”

Release Violations/Fugitive

Traffic Violations

Narcotics

THEFT

Other Crimes

##   year Simple.Assault Release.Violations.Fugitive Traffic.Violations Narcotics
## 1 2016            232                          94                171       171
## 2 2017            217                         123                137       137
## 3 2018            191                          93                129       129
## 4 2019            212                          84                112       112
## 5 2020            146                          58                 42        42
## 6 2021            114                          30                 47        47
##   Theft Other.Crimes
## 1    57           63
## 2    71           88
## 3    60           53
## 4    59           75
## 5    20           34
## 6    21           25

Some plots of crime trend of different generation

## 
## 载入程辑包:'gridExtra'
## The following object is masked from 'package:dplyr':
## 
##     combine

Barplot of crime trend of white males of different generations

Plot each crime separately:

Plot all 6 crimes together but with different age groups

Though the samples of three generations (young middle age, old people) are different, the behaviors of each crime trend plots are all the same. There is no obvious difference among three groups, thus we can see “Age” do not have a significant effect on the trend of these common crimes.

Analysis

Since crime is likely to be a rare event, the number of occurrences per day of a given crime is expected to follow Poisson distribution. Poisson distribution is a distribution used to describe the distribution of the number of rare phenomena when a large number of them are observed. If a distribution follows Poisson distribution, and the average number of occurrences of the phenomenon is \(\lambda\), the probability that the phenomenon will occur \(x\) times is given by \[p(x) = \exp(-\lambda)\frac{\lambda^{x}}{x!}.\] In the following, we will estimate \(\lambda\) of each crime before and after COVID-19 to see there is a difference in crime trend.

Offenses Against Family & Children

Before COVID-19

The trend of “Offenses Against Family & Children,” Domestic Violence (DV), appears to have changed after COVID-19. The frequency table of DV before COVID-19 is as follows.

# of occurrences per day Frequency Relative frequency
0 1364 0.9336071
1 95 0.065024
2 2 0.0013689
3 0 0

We can calculate \(\lambda\) from the above table and \(\lambda = 0.0678\). We will plot the histogram and Poisson distribution with \(\lambda = 0.0678\) to check if they match or not.

We can see that the Poisson distribution fits well with the histogram.

Next, we try to estimate \(99\%\) Confidence Interval of \(\lambda\). The variance of Poisson distribution is equal to its mean (\(\lambda\)). Therefore, \(99\%\) Confidence Interval of \(\lambda\) can be written as \[ \bar{x} - z_{*}\cdot\sqrt{\frac{\bar{x}}{n}} \leq \lambda \leq \bar{x} + z_{*}\cdot\sqrt{\frac{\bar{x}}{n}}, \] where \(\bar{x}\) is the sample mean, \(n\) is the sample size, and \(z_*\) is z-value corresponding to the \(99\%\) confidence interval, and the value is 2.58. From this expression, 99% Confidence Interval of \(\lambda\) for DV before COVID-19 is [0.05, 0.0856].

After COVID-19

The frequency table of DV after COVID-19 is as follows.

# of occurrences per day Frequency Relative frequency
0 680 0.9302326
1 47 0.0642955
2 1 0.001368
3 0 0
4 1 0.001368
5 0 0
0 0
44 0 0
45 1 0.001368
46 0 0
0 0
77 0 0
78 1 0.001368
79 0 0
0 0

There are two outliers (45 and 78) in the table. The dates of them are 1 and 1. Since these dates are correspond to “Capitol attack” and “George Floyd protests”, we will drop the value of these dates.

The calculated \(\lambda = 0.0725\). The histogram and the poisson distribution with \(\lambda = 0.0725\) are shown in Figure 16.

The Poisson distribution fits well with the histogram.

99% Confidence Interval of \(\lambda\) for DV after COVID-19 is [0.0465, 0.0985].

Comparing Confidence Intervals

Figure 17 shows the Confidence Intervals before and after COVID-19. There was overlap in the Confidence Intervals, and it is not possible to say that there was a change in the \(\lambda\) of “Offenses Against Family & Children” before or after COVID-19.

Traffic Violations

Before COVID-19

The trend of “Traffic Violations” also appears to have changed after COVID-19. The frequency table of Traffic Violations before COVID-19 is as follows.

# of occurrences per day Frequency Relative frequency
0 602 0.4123288
1 530 0.3627652
2 225 0.1540041
3 77 0.0527036
4 22 0.0150582
5 4 0.0027379
6 1 6.844627^{-4}
7 0 0

The calculated \(\lambda = 0.907\). The histogram and the poisson distribution with \(\lambda = 0.907\) are shown in Figure 19.

The Poisson distribution fits well with the histogram.

99% Confidence Interval of \(\lambda\) for Traffic Violations before COVID-19 is [0.842, 0.972].

After COVID-19

The frequency table of Traffic Violations after COVID-19 is as follows.

# of occurrences per day Frequency Relative frequency
0 546 0.746922
1 156 0.2134063
2 23 0.0314637
3 3 0.004104
4 3 0.004104
5 0 0

The calculated \(\lambda = 0.306\). The histogram and the poisson distribution with \(\lambda = 0.306\) are shown in Figure 21.

The Poisson distribution fits well with the histogram.

99% Confidence Interval of \(\lambda\) for Traffic Violations before COVID-19 is [0.252, 0.358].

Comparing Confidence Intervals

Figure 22 shows the Confidence Intervals before and after COVID-19. There was no overlap in the Confidence Intervals, and there may have been a change in the Traffic Violations lambda before and after COVID-19.

Simple Assault

Before COVID-19

The below table shows the frequency and relative frequency of Simple Assault before COVID-19.

# of occurrences per day Frequency Relative frequency
0 438 0.2997947
1 479 0.3278576
2 284 0.1943874
3 156 0.1067762
4 76 0.0520192
5 13 0.008898
6 8 0.0054757
7 4 0.0027379
8 1 6.844627^{-4}
9 2 0.0013689
10 0 0

We got \(\lambda = 1.36\) by calculating the average of occurrences per day.

After COVID-19

The frequency and relative frequency in 2020 and 2021 is shown in below. The \(\lambda\) for 2020 and 2021 was \(0.923\).

# of occurrences per day Frequency Relative frequency
0 297 0.4062927
1 267 0.3652531
2 112 0.1532148
3 43 0.0588235
4 8 0.0109439
5 3 0.004104
6 0 0
7 0 0
8 1 0.001368
9 0 0

Comparing Confidence Intervals

Figure 26 shows the Confidence Intervals before and after COVID-19. There was no overlap in the Confidence Intervals, and there may have been a change in the Simple Assault lambda before and after COVID-19.

Theft

Before COVID-19

The frequency and relative frequency in 2016 to 2019 is shown in below. The \(\lambda\) before COVID-19 was \(0.404\).

# of occurrences per day Frequency Relative frequency
0 973 0.6659822
1 398 0.2724162
2 79 0.0540726
3 10 0.0068446
4 1 6.844627^{-4}
5 0 0

After COVID-19

The frequency and relative frequency in 2020 and 2021 are shown in below. The \(\lambda\) for 2020 and 2021 was \(0.115\).

# of occurrences per day Frequency Relative frequency
0 653 0.8932969
1 72 0.0984952
2 6 0.0082079
3 0 0

Comparing Confidence Intervals

Figure 31 shows the Confidence Intervals before and after COVID-19. There was no overlap in the Confidence Intervals, and there may have been a change in the Theft lambda before and after COVID-19.

Interpretation

Simple Assault and Traffic Violations

Statistically significant reductions in Simple Assault and Traffic Violations were observed for \(\lambda\) before and after COVID-19. Since these crimes seem to be more likely to occur the more people are out, it is likely that the restrictions and curbs on going out due to COVID-19 contributed to the decrease in these crimes.

Theft

A statistically significant decrease in theft was also observed in \(\lambda\) before and after Corona. Considering that thefts are committed against empty homes, the decrease in empty homes due to the curfew restrictions caused by COVID-19 may have contributed to the decrease in thefts.

Offenses Against Family & Children

The more time one spends at home due, the more Offenses Against Family & Children are likely to increase. In fact, in terms of the number of cases alone, Offenses Against Family & Children have increased after COVID-19. At first glance, the curfew restrictions caused by COVID-19 seems to be the cause. However, most of these cases were caused by special incidents unrelated to COVID-19, and when these effects were removed, there was no statistically significant difference in the change in Offenses Against Family & Children before and after COVID-19. As for white males in the DC area, Offenses Against Family & Children to the point of arrest does not appear to be affected by the changes in their lives caused by COVID-19.

Appendix - Other Major Crimes

Release Violations/Fugitive

Before COVID-19

# of occurrences per day Frequency Relative frequency
0 817 0.559206
1 428 0.29295
2 168 0.1149897
3 40 0.0273785
4 5 0.0034223
5 3 0.0020534
6 0 0

After COVID-19

# of occurrences per day Frequency Relative frequency
0 549 0.751026
1 154 0.2106703
2 24 0.0328317
3 4 0.005472
4 0 0

Comparing Confidence Intervals

Driving/Boating While Intoxicated

Before COVID-19

# of occurrences per day Frequency Relative frequency
0 875 0.5989049
1 416 0.2847365
2 138 0.0944559
3 25 0.0171116
4 6 0.0041068
5 1 6.844627^{-4}
6 0 0

After COVID-19

# of occurrences per day Frequency Relative frequency
0 527 0.7209302
1 168 0.2298222
2 29 0.0396717
3 5 0.0068399
4 2 0.002736
5 0 0

Comparing Confidence Intervals

Narcotics

Before COVID-19

# of occurrences per day Frequency Relative frequency
0 1053 0.7207392
1 318 0.2176591
2 66 0.0451745
3 10 0.0068446
4 5 0.0034223
5 6 0.0041068
6 0 0
7 1 6.844627^{-4}
8 0 0
9 1 6.844627^{-4}
10 0 0
11 0 0
12 0 0
13 1 6.844627^{-4}
14 0 0

After COVID-19

# of occurrences per day Frequency Relative frequency
0 652 0.8919289
1 66 0.0902873
2 10 0.0136799
3 3 0.004104
4 0 0

Comparing Confidence Intervals

Liquor Law Violations

Before COVID-19

# of occurrences per day Frequency Relative frequency
0 1090 0.7460643
1 259 0.1772758
2 76 0.0520192
3 32 0.0219028
4 3 0.0020534
5 1 6.844627^{-4}
6 0 0

After COVID-19

# of occurrences per day Frequency Relative frequency
0 699 0.9562244
1 27 0.0369357
2 4 0.005472
3 1 0.001368
4 0 0

Comparing Confidence Intervals